The report of dreams is subjective (prone to be based on feeling rather than actual fact). It would be more scientific to back up the subject's self-report with objective data (precise measurement). In this study, dreams are to be related to physiological phenomena (physical measurable processes within the body). There were two physiological measurements made.
Previous findings have demonstrated:
Three tests were carried out:
7 males
2 females
All were adults.
5 were studied intensively. The other four were only tested on one or two nights. This would give some indication as to whether the observed effects were partly owing to the subjects having their sleep disrupted over many nights (between 6 and 17 nights, with between 50 and 77 awakenings).
Subjects reported to a sleep lab at their normal bedtime. They had abstained from alcohol and caffeine on the day of the experiment. The subjects were asked to relay to the experimenters, by way of a microphone, the required information about the content of their dream and sometimes the length of dream. This was to be done as soon as they were woken by a doorbell, without the experimenter talking to them. Occasionally the experimenter would try to clarify some detail of the dream by entering the room and talking directly to the subject.
The subjects normally fell back to sleep five minutes after being woken. The average number of awakenings in one night were 5.7, and the average sleeping time was 6 hours
All subjects displayed REM at various times of the night and the REM was accompanied by a low-voltage fast EEG pattern. In-between the REM periods the EEG pattern was high-voltage and slow. This period is called NREM (no rapid eye movements). At the beginning of sleep the subject would, however, pass through a period of fast EEG without REM.
REM periods ranged from 3 to 50 minutes, with an average of 20 minutes. REM periods became longer as the night progressed.
The REM was not continuous, but appeared in bursts of between 1 and 100 very rapid movements. Each movement took about 0.15 second. On average a REM period would occur once every 92 minutes.
Dement and Kleitman maintain that the multiple awakenings did not disrupt these patterns, as similar patterns had been found in a previous experiment (Dement & Kleitman 1955).
If the patient was woken during a REM period, subjects would usually report a dream. Once woken during REM, dreaming was not resumed until the next REM period. The exception to this was towards the end of sleep, when subjects would sometimes continue with their dreaming after being woken. REM periods were found to be relatively long at this time.
The subjects were woken either in REM or NREM sleep. 2 subjects were woken randomly. 1 subject was woken 3 times in REM then 3 times in NREM. 1 subject was told he would only be woken in REM periods, but was woken randomly. 1 subject was woken whenever the experimenter wanted to!
As can be seen below when subjects were woken during REM they tended to recall dreams, but when woken during NREM they often could not.
The subject, who was mislead into believing that he would only be woken during REM, was not affected by this deception.
To check for practice effects the subjects' performances during the first half of the trials were compared to their performances in the second half of the trials. There was no evidence to suggest that the subjects got better at recalling their dreams.
With 17 awakenings, 5 dreams were recalled within 8 minutes of entering NREM. This could be because the subject still remembered a dream that had actually occurred during REM sleep. Only 6 dreams were recalled in 132 awakenings that occurred after 8 minutes of NREM sleep.
Most of the failures to recall a dream, during REM, occurred earlier in the night.
Subjects were woken at various times during REM but were not able to estimate the length of their dreams to the nearest minute. To make the task easier the subject was asked to say whether he or she had been dreaming for either 5 or 15 minutes. They were woken randomly at one of these two times. The Pie charts below demonstrate how accurate the subjects were at estimating the length of their dreams.
In addition the length of the subjects' narrative, recalling the dreams was correlated with the length of the REM periods. Significant positive correlations were found for all five of the main subjects.
Subjects were woken during one of four eye-movement patterns.
3 out of 35 awakenings were for vertical movements. Subjects reported standing at a bottom of a cliff hoisting things up and down, climbing a ladder whilst looking up and down, and looking down to pick up a basketball followed by looking up to throw the ball through the basket.
There was one horizontal awakening. The subject reported watching 2 people throw tomatoes at each other.
10 awakenings during little or no eye movement found subjects dreaming about looking at things in the distance. Two of these subjects displayed an eye-movement to the left shortly before being woken. Both were dreaming about driving their car, when either a car came speeding at the subject from the left, or a passer-by hailed the subject from the left.
21 awakenings occurred during the mixture of eye-movements. Subjects tended to be observing people or objects close to them.
The eye movements produced by the sleeping subjects were compared with wide-awake subjects viewing distant or close-by objects, and similar patterns were noted.